Winter Storm of 15 December 2005 By Richard H. Grumm National Weather Service Office State College, PA 16803 1. INTRODUCTION A complex winter storm brought snow, sleet, and freezing rain to central Pennsylvania. Warm air penetrated aloft early in the event which limited snowfall totals in southern areas of the State where the snow rapidly changed to sleet. The low-level cold air slowly retreated and most southern areas eventually turned to rain, after a period of freezing rain. Snowfall ranged from around 7 inches along the New York border to around 5 inches in State College, and 0.5 to 1 inches in southernmost regions of the State. Few locations met the National Weather Service criteria for heavy snow in central Pennsylvania, defined as 6 inches of snow in 12 or 24 hours. However, the 1 inch of sleet 1 and accumulating ice met the winter storm criteria over a wide area of central Pennsylvania. This event was well forecast by the National Centers for Environmental Predictions (NCEP) ensemble predictions systems (EPS) forecast this potential winter storm quite accurately several days in advance. Precipitation type forecast in the Short-Range ensemble forecast system (SREF) showed the uncertainty in the precipitation type at all central Pennsylvania forecast sites. These forecasts will be shown to demonstrate 1 Personal observation of 1.1 inches of sleet between 1900 to 2300L 15 December 2005. the utility of using ensemble forecasts in complex winter storms. Overall, this event was well forecast by the NCEP EPS s. The uncertainty in the precipitation type forecasts was well predicted by the SREF EPS. On 14 December, a major ice storm affected southern Virginia and North Carolina. In Pennsylvania, snow and sleet dominated early in the event and most locations only saw a few hours of freezing rain, reducing the damaging impacts of this storm. This paper will document the winter storm of 15 December 2005. The utility of EPS data in the winter storm forecast process is also presented. 2. METHODS Most data used here were obtained in real-time and were based on products from the NCEP SREFs that were available during the event. The focus will be on SREF precipitation type products such as EPS plume diagrams. Other key SREF fields are shown to demonstrate how the EPS predicted the overall event. All SREF data are plotted using GrADS (Doty and Kinter 1992). Climatic anomalies are produced from the SREF as described by Grumm and Hart (2001a) and Hart and Grumm (2001). Applications of anomalies to cold season winter storms were documented by Grumm and Hart (2001a and 2001b).
Figure 1 shows the SREF MSLP forecasts initialized at 13/0900 UTC valid at 15/1200 UTC. These data show the northern stream cyclone moving into the Great Lakes and the southern stream cyclone over the northern Gulf of Mexico. Above normal surface pressure is over the northeastern United States associated with the surface anticyclone that brought record low temperatures to the region on the 14 th. The southern stream low was forecast to be over the Delmarva by 16/1200 UTC (not shown). Figure 1 SREF MSLP forecasts initialized at 0900 UTC 13 December 2005 valid at 1200 UTC 15 December showing a) spaghetti plot and spread b) ensemble mean and the departure from normal in standard deviations from normal. Data depicted here will have times referred to as day and hour. Therefore, a forecast initialized at 0900 UTC 15 December 2005 may be referred to as 16/0900 UTC. It would be prohibitive to show a full set of forecasts from each EPS and each forecast cycle. A limited sample of EPS output related to the forecast problem is presented here. To keep the size down, different products are shown at varying forecast ranges to show the variety of potential products available. Those not shown for a particular time period may be referred to. 3. RESULTS i. 13 December forecasts The shallow cold air was in the SREF s and in the warm advection ahead of the southern stream cyclone; the SREFs forecast a prolonged period of freezing rain in the southeastern United States. The type of precipitation forecast at the mid-point of the time of maximum freezing rain is shown in Figure 2. There was a high probability of freezing rain over western North Carolina. The SREFs forecast the time from 15/0000 to about 15/2100 UTC as a high probability time for freezing rain over North Carolina. Farther north over Pennsylvania, the SREFS forecast snow, sleet, and freezing rain. In southern areas, the forecasts, as shown on plume diagrams, showed the high probability of above freezing temperatures and rain after 16/0300 UTC. SREF forecasts initialized at 13/2100 UTC continued to show the same general forecasts. A potential ice storm in the western Carolina s and Virginia and a snow to mixed precipitation event in Pennsylvania. The plume diagrams showing 2m Temperatures and precipitation type
Figure 2 SREF initialized at 0900 UTC 13 December 2005 valid at 0900 UTC 15 December 2005. Each panel show the probability of categorical precipitation type and the 3-hour precipitation amounts. clearly show this forecast evolution (Figure 3). The plume for State College suggested snow or ice pellets at onset of precipitation. Some members forecast up to 0.60 inches of total liquid falling as snow. The mean snowfall was forecast at 2.5 inches (10:1 first guess). The indications also were for some significant freezing rain and sleet potential. The precipitation type (PTYPE) forecasts clearly indicated a lot of uncertainty in these forecasts. The 2m temperatures, showing the near record lows on the 14 th and gradual warm up implied a very cold boundary layer which would favor the frozen to freezing scenario. One problem with the forecasts (not shown) was that some of the operational GFS runs indicated all snow at State College. Forecasts at Harrisburg showed less frozen and freezing precipitation and a significant number of members forecast a period of rain. The freezing rain forecasts implied potential for significant ice accumulations. ii. 14 December forecasts As will be shown, forecasts from the 14 th did little to clear up the matter of what form the precipitation would fall as. The PTYPE forecasts continued to show the high probability of freezing rain in the western Carolina s and Virginia (not shown). Figure 5 shows the end of the ice event and the potential for snow, freezing rain, sleet, and rain over Pennsylvania. The high probability of a rain was focused along the coastal plain.
Figure 3 SREF forecasts initialized at 2100 UTC 13 December showing (upper) accumulated precipitation color coded by type and 3-hour instantaneous precipitation and (lower) 2m temperature forecasts by each EPS member. Thick black line is the ensemble mean. Figures 6 & 7 show the 850 hpa winds and 850 hpa temperatures as forecast from the 0900 UTC 14 December SREF valid at 0000 UTC 16 December. The strong and anomalous southerly (southsoutheasterly) flow over the low-level cold air is clearly evident in Figure 6. The 850 hpa temperatures show the cold air east of the mountains and a wedge of warm air moving into the Ohio Valley. The lack of a significant easterly jet, a key feature in most east coast winter storms was a missing feature in these data. Forecasts from 2100 UTC 14 December continued to show a high probability of a mixed precipitation event. Plumes and precipitation type forecasts (not shown)
Figure 4 As in Figure 3 except PTYPE fore Harrisburg, PA. Figure 9 shows the ensemble forecasts of MSLP. The two cyclones, both with below normal surface pressures are present. The strong gradient on the north side of these systems implies a strong southeasterly jet and relates well to the PWAT anomaly field in Figure 8. The ridge of high pressure with above normal surface pressure shows the low-level cold air was slow to retreat. suggested snow to ice and freezing rain in central sections of the state and snow to freezing rain and eventually all rain in the eastern and southeaster sections of the state. Figure 8 shows the precipitable water forecasts valid at 16/0000 UTC from the SREF initialized at 14/2100 UTC. The moisture fields show the conveyer belt of moist air over the cold air and a surge of above normal precipitable water (PWAT) in the northern stream low. The highest PWAT anomalies were along the southeastern United States associated with the southern stream low. Studies of previous ice storms, such as the January 1998 event showed the association of high precipitiable water over the frontal boundary. iii. 15 December forecasts The SREF forecasts initialized at 0900 UTC 15 December showed more snow and ice in central Pennsylvania and less rain than previous forecasts. This is demonstrated by the plume for State Figure 5 As in Figure 2 except SREF forecasts initialized at 0900 14 December 2005 valid at 0300 UTC 16 December 2005.
Figure 6. SREF forecasts of 850 hpa winds initialized at 0900 UTC 14 December 2005 valid at 0000 UTC 16 December 2005. Upper panel shows 850 hpa winds and U-wind anomalies and lower panels shows 850 hpa winds and V-anomalies. Figure 7. SREF forecasts initialized at 0900 UTC 14 December 2005 valid at 0000 UTC 16 December 2005. Upper panel shows spaghetti plot of select 850 hpa isotherms and dispersion about the mean and lower panels shows the ensemble mean 850 hpa temperature field and anomalies. College (Figure 10). For comparison purposes, the plume from 24 hours earlier is also shown. The earlier forecast showed several members forecasting rain. There was no rain in the 15/0900 UTC forecasts at State College. Both forecasts showed snow, however the mean was 0.31 with a maximum of 0.54 liquid equivalents as snow in the more recent forecasts. Forecasts from the 14 th implied the potential, in an all snow scenario, for up to 10 inches of snow. The plume diagram (Figure 10) also showed a convergence on the timing of the precipitation. Note the strong clustering for all members to start the precipitation after 1500 UTC on the 15 th and end it around 1200 UTC on the 16 th. As the event drew nearer, the uncertainty decreased. Figure 11 shows the precipitation type plume and 2m temperature plume for a point near Harrisburg. These data show the cold air rapidly losing their grip on the region and a period of sleet and ice then rain. Harrisburg has less than an inch of snow and sleet, a brief period of freezing rain and then changed to rain. Surface temperatures actually rose into the 40s over much of the lower Susquehanna valley overnight. 4. CONCLUSIONS A complex winter storm affected the eastern United States on 15 and 16 December 2005. This storm brought
significant ice accumulations, causing widespread power outages in the western Carolina s and Virginia. In Pennsylvania, the event produced snow, sleet, and freezing rain. Eastern and southeastern parts of the state saw rain. The sleet and freezing rain limited snowfall accumulations. The SREF precipitation type forecasts did a remarkable job forecasting the North Carolina ice storm and the mixed precipitation event that affected Pennsylvania. From a forecast perspective, the uncertainty in the snow, sleet, and ice amounts displayed in the forecasts was realized in the observations. For example, at State College, the SREFS showed a potential for 4 to 13 inches of snow at various times from forecasts initialized between 13/0900 and 15/0900 UTC. These precipitation type forecasts also showed the potential for an inch of sleet and some freezing rain. The shorter duration forecasts more accurately mimicked the observed conditions. A few members showed a period of rain. In and around State College 3 to 5 inches of snow was observed. An inch of sleet fell over the region and a period of freezing rain was observed. Thus 3 of the 4 forecast precipitation types forecast were observed. Figure 8. SREF forecasts initialized at 2100 UTC 14 December 2005 valid at 0000 UTC 16 December 2005. Upper panel shows spaghetti plot of select precipitable water isotherms (mmm) and dispersion about the mean and lower panels shows the ensemble mean precipitable water field and its departure from normal in standard deviations.. Figure 9 same as Figure 8 except MSLP forecasts.
Figure 10 SREF plume diagrams for a point near State College from SREFs initialized at 0900UT 14 and 15 December 2005. Overall, the SREF 2m temperature and precipitation types reflected the character of the storm. They also reflected the uncertainty in forecasting this storm. The exact conditions of the winter storm for warning purposes was not well forecast by the SREF until about 10 hours before the onset of precipitation over Pennsylvania. The 0900 UTC 15 December SREFs proved to be a valuable short-term forecast tool.
Figure 11. SREF forecasts initialized at 1500 UTC showing precipitation type and 2 meter temperature forecasts for a point near Harrisburg, Pennsylvania. Meteorologically, the SRFS contained signals that did not favor a major snow event. The strong southeasterly jet, with mainly south-southeasterly winds at 850 hpa is a common characteristic of rain and mixed precipitation events. Most significant east coast winters snow storms have a low-level easterly jet which this event lacked. This suggested a low probability of an all snow event, favoring a mixed or snow to rain event. Previous examples using climatic anomalies have shown the association of high precipitiable water over the shallow frontal boundaries as significant
science visualization. International Conference on Interactive Information and Procession Systems, Atlanta, Georgia, 5-10 January, 1992. Grumm, R.H., and RE Hart. 2001a: Standardized Anomalies Applied to Significant Cold Season Weather Events: Preliminary Findings. Weather and Forecasting,16, 736 754. Grumm, R. H., and Hart R., 2001b: Anticipating heavy rainfall events: Forecast aspects. Preprints, Symp. on Precipitation Extremes: Prediction, Impacts, and Responses, Albuquerque, NM, Amer. Meteor. Soc., 66 70. ingredients in ice storms. This case at times was forecast and observed to have this characteristic. Hart, R.E., and R H. Grumm. 2001: Using Normalized Climatological Anomalies to Rank Synoptic-Scale Events Objectively. Monthly Weather Review,129,2426 2442. 5. ACKNOWLEDGEMENTS A special thanks to NCEP for access to the SREF data in real-time. 6. REFERENCES Doty, B. and J.L. Kinter III, 1992: The Grid Analysis and Display System (GrADS): A practical tool for research